Switch to upstream cassandra-all and fix issues with vpc internal clusters

(Bugfix) Switch to using upstream cassandra-all from our internal jar
(Bugfix) Allow Priam to run on clusters that don't have public-hostnames and ips

BackupNotifications filter

(#844, #839) Implementation of a filter for Backup Notification. The filter can be controlled using the configuration "priam.backupNotifyComponentIncludeList"

Backup notifications filter implementation

(#842, #839) Implementation of a filter for Backup Notification. The filter can be controlled using the configuration "priam.backupNotifyComponentIncludeList"

Transitive dependency fix

Adding google to the list of repositories to resolve google related transitive dependencies.

Move flush and compactions to Service. Send SNAPSHOT_VERIFIED message once a snapshot is verified.

(#836) Move flush and compactions to Service Layer. This allows us to "hot" reload the jobs when configurations change.
(#836) Send SNAPSHOT_VERIFIED message when a snapshot is verified and ready to be consumed by downward dependencies.

Move flush and compactions to Service. Send SNAPSHOT_VERIFIED message once a snapshot is verified.

• (#835) Move flush and compactions to Service Layer. This allows us to "hot" reload the jobs when configurations change.
• (#835) Send SNAPSHOT_VERIFIED message when a snapshot is verified and ready to be consumed by downward dependencies.

Fix X-Y-Z and other bug fixes

(#834 ) Removing functionality of creating incremental manifest file in backup V1 as it is not used.
(#834 ) Bug fix: When meta file do not exist for TTL in backup v2 we should not be throwing NPE.
(#834 ) Bug fix: Fix X-Y-Z issue using gossip status information instead of gossip state. Note that gossip status is (JOINING/LEAVING/NORMAL) while gossip state is (UP/DOWN). Gossip state is calculated individually by all the Cassandra instances using gossip status.Priam only checked the database to give to Cassandra when "replacing" a token. That ip could be old or outdate. Now the replacing Priam asks other Priam for the view as from C* (source of truth) to find the IP address to replace.
(#834 ) Bug fix: Change return type of cassadmin/paritioner from JSON to text type.
(#834 ) Bug fix: Handle files of 0 bytes iff created from C* during snapshot operation.
(#834 ) Bug fix: Reduce the no. of max chunks as files could inflate after compression.

Migrate to openjdk8 for Travis

(#832) Travis build fails for oraclejdk8. Migration to openjdk8

Fix X-Y-Z and NPE while running TTL on backup V2.

(#827) Removing functionality of creating incremental manifest file in backup V1 as it is not used.
(#827) Bug fix: When meta file do not exist for TTL in backup v2 we should not be throwing NPE.
(#827) Bug fix: Fix X-Y-Z issue using gossip status information instead of gossip state. Note that gossip status is (JOINING/LEAVING/NORMAL) while gossip state is (UP/DOWN). Gossip state is calculated individually by all the Cassandra instances using gossip status.

Bug fix: Rollback fix to use Gossip info while grabbing dead and pre-assigned tokens.

(#826): Rollback the fixes to use Gossip info while grabbing dead and pre-assigned tokens. Gossip info doesn't not reflect the correct cluster state always. A node marked with status as NORMAL in the Gossip info could actually be down. This can be checked using nt ring. This change will unblock the nodes from joining the ring.